Cat Jam stealth coding model is free on Vercel AI Gateway with a 1M-token context window
Cat Jam, a new stealth reasoning model on Vercel AI Gateway, is free to use during its stealth period. Call it as stealth/cat-jam. It is built for coding and agentic tasks, supports function calling and structured outputs, has a 1M-token context window, and works in Claude Code, Codex, Cursor, and other coding agents.
The briefing
Vercel AI Gateway added Cat Jam on October 15, 2026. It is an anonymous stealth model, so the provider has not been named yet. It is a low-latency reasoning model for coding and agentic tasks, and calls are free during the stealth period.
What you can use
Input and output for Cat Jam on AI Gateway are free while the model is in stealth. No end date for the stealth period has been announced.
The model ID is stealth/cat-jam, with a 1M-token context window and up to about 131K output tokens. It supports function calling, so agents can connect it to your own tools, and structured outputs that follow a JSON schema you supply. Responses include the model's reasoning text alongside the answer. Reasoning effort can be set to low, medium, high, xhigh, or max.
It accepts text input only. Zero Data Retention (ZDR) is not available, and prompts and responses may be used for training and model improvement.
How to use it
- Sign in to Vercel and create an AI Gateway API key for your team.
- Call it through the AI SDK, the OpenAI-compatible Chat Completions and Responses APIs, or the Anthropic-compatible Messages API, with model
stealth/cat-jam. - To use it in Claude Code, Codex, or Cursor, install the latest Vercel CLI, run
npx vercel ai-gateway setup, then pick stealth/cat-jam in your agent.
Official pages: Vercel changelog · Cat Jam model page.